Median Unbiased Estimation of Bivariate Predictive Regression Models with Heavy-tailed or Heteroscedastic Errors

In this paper, we consider median unbiased estimation of bivariate predictive regression models with non-normal, heavy-tailed or heteroscedastic errors. We construct confidence intervals and median unbiased estimator for the parameter of interest. We show that the proposed estimator has better predictive potential than the usual least squares estimator via simulation. An empirical application to finance is given. And a possible extension of the estimation procedure to cointegration models is also described.

The Effects of Demographic Features on Differences in Sensitivity between PCL-C and SCL-90 Scores in a Follow-up Study in Secondary School Students in the Wenchuan Earthquake Region

Objective To analyze the sensitivity of effect factors between the PCL-C and the SCL-90, to provide evidence for social psychological crisis screening and post-trauma interventions. Methods We administered the PCL-C and SCL-90 to screen for PTSD and other psychological problems among students who survived the disaster and continued their school studies. The surveys were carried out 3, 6, 9, and 12 months after the earthquake. A bivariate 2-level logistic model was used to explore the different levels of sensitivity among students. The factors influencing the relationships between PTSD and depression, and between PTSD and anxiety were examined. Results We analyzed data from 1677 students, revealing that female students in higher grades were more likely to exhibit symptoms of depression, rather than PTSD, compared with the control group （males in lower grades）, and the difference was significant （P〈0.05）. In contrast, ethnic minorities were more likely to exhibit PTSD symptoms compared to the others. In addition, female students were more likely to exhibit symptoms of anxiety than PTSD. Other effects that did not reach statistical significance were suggested to have a similar influence on PTSD, depression, and anxiety. Conclusion After a natural disaster, specific aspects of depression and anxiety should be examined, avoiding an overemphasis on PTSD in social psychological crisis interventions.

Exploration of the contributing factors to the walking and biking travel frequency using multi-level joint models with endogeneity

The enormous advantages of active transportation lead the transportation research focus towards enhancing the walking and biking trips.The present study explored the influential factors to the walking and biking travel frequency based on data obtained from the National Household Travel Survey California add-on survey.The study features some highlights.First,bivariate models were used to account for the common unobserved heterogeneity shared at both household and personal levels.Second,endogeneity was explicitly considered.Third,both variable importance ranking and correlation analysis are employed to determine the different features to be fed into each of the joint models.The results illustrated that the models developed with endogeneity performed better than the models without endogeneity.Four influential variables which includes mode to work by bicycle,public transit usage,count of household members,and multiple race responses,tend to have statistically significant impacts on walking and biking trips.

Bivariate macro-level safety analysis of non-motorized vehicle crashes and crash-involved road users

The high risk of injury resulting from non-motorized vehicle(NMV)crashes has created the goal of using the 3E strategy to comprehensively improve NMV safety.Traditional safety improvement methods identify hot zones generally by crash frequency or density,which is effective for roadway engineering improvements but neglects characteristics related to other improvements such as safety education.As safety education would be more effective if targeted at the residences of crash-involved road users,the traditional approach to hot zones may therefore provide biased results for such alternative countermeasures.After confirming that 77.17%of NMV crashes occurred outside the involved riders’areas of residence,this study compared the differences between the locations of crashes and the residences of NMV crash-involved riders in safety influencing factors and hot zone identification.A Poisson lognormal bivariate conditional autoregressive(PLN-BCAR)model was developed to account for potential correlations between crashes and involved riders.The model was compared with the univariate Poisson lognormal conditional autoregressive(UPLN-CAR)model and the bivariate Poisson lognormal conditional autoregressive(BPLNCAR)model;the PLN-BCAR model outperformed the other two models in its better interpretation of the influencing factors and its more efficient parameter estimation.Model results indicated that crashes were mainly associated with roadway and land use characteristics,while involved road users were mainly associated with socioeconomic and land use characteristics.The potential for safety improvement method was adopted to identify hot zones for countermeasure implementation.Results showed over 60%of the identified hot zones were inconsistent:they needed improvement in either engineering or education but not both.These findings can help target the type of improvement to better utilize resources for NMV safety.

Exploring the factors affecting electric bicycle riders'working conditions and crash involvement in Ningbo,China

The rapid development of the delivery industry brings convenience to modern society.However,the high rates of crashes and the survival of electric bicycle(e-bike)riders in the delivery industry raise concerns.The primary objective of this study is to explore the factors affecting delivery e-bike riders’stressful work pressure and crash involvement in China.Data were collected by a questionnaire survey administered in the city of Ningbo,China.A bivariate ordered probit(BOP)model was developed to simultaneously examine the factors associated with both the working conditions of delivery e-bike riders and their involvement in crashes.The marginal effects for the contributory factors were calculated to quantify their impacts on the outcomes.The results showed that the BOP model can account for commonly unobserved characteristics of the working conditions and crash involvement of delivery e-bike riders.The BOP model results showed that the stressful working conditions of delivery e-bike riders were affected by the number of orders and delivery time and rider age and risky riding behaviors.Delivery rider involvement in crashes was affected by the number of orders,strength of the punishment for traffic violations,and familiarity with traffic regulations.It was also found that stressful working conditions and crash involvement were strongly and positively correlated.The findings of this study can enhance our understanding of the factors that affect the working conditions and delivery rider crash involvement.Based on the results,some suggestions regarding public policy,risky riding behaviors,safety promotion,and stronger corporate governance rules were discussed to increase the targeted safety-related interventions for delivery ebike riders in Ningbo,China.

Covering or monitoring?Independent director connectedness and corporate fraud in China

This paper examines how independent directors’social capital,as measured by their social network,affects corporate fraud.We find that firms with wellconnected independent directors are less likely to commit fraud,supporting our monitoring effect hypothesis.This result is robust to a battery of tests.Further analyses show that the effect is stronger for firms with a relatively poor legal environment,for firms whose independent directors face strong reputation incentives and when independent directors are audit committee members.Moreover,we explore a potential economic mechanism of the effect and observe that well-connected independent directors are associated with less absenteeism and more dissension.Overall,our findings suggest that independent directors’social capital plays an important role in corporate governance.

How has COVID-19 changed trip patterns by purpose in China?

COVID-19 has upended the whole world. Due to travel restrictions by governments and increased perceived risks of the disease, therehave been significant changes in social activities and travel patterns. This paper investigates the effects of COVID-19 on changes toindividuals’ travel patterns, particularly for travel purposes. An online questionnaire survey was conducted in China, which incorporatesquestions about individuals’ sociodemographic and travel characteristics in three different periods of COVID-19 (i.e. before theoutbreak, at the peak and after the peak;the peak here refers to the peak of the pandemic in China, between the end of January and1 May, 2020). The results show that trip frequency decreased sharply from the outbreak until the peak, and drastically increased afterthe peak. Nevertheless, the data fromthis study suggests that it has not fully recovered to the level before the outbreak. Subsequently,a series of random parameters bivariate Probit models for changes in travel patterns were estimated with personal characteristics.The findings demonstrate that during the peak of the pandemic, residents who did not live in more developed cities reached lowfrequencytravel patterns more quickly. For travel purposes, residents of Wuhan, China resumed travelling for work, entertainmentand buy necessities at a much higher rate than other cities. After the peak, students’ travel for work, entertainment and to buy necessitiesrecovered significantly faster than for other occupations. The findings would be helpful for establishing effective policies tocontrol individual travel and minimize disease spread in a possible future pandemic.

Patent Application, Grant and Patent Law Amendment in China

This study employs 920,108 invention patents from the State Intellectual Property of China(SIPO)to examine the impact of two patent law amendments,which occurred in 1993 and 2001 respectively,on the patent grant.The first patent law amendment in 1993 led to less strict criteria for patent approval.And the second patent law amendment,which occurred in 2001,showed a similar positive impact on patenting as well.It is more likely that the first patent law amendment encouraged foreign applicants to make patent application,because their applications and grant share started to rise after 1993.By contrast,the second patent law amendment played an opposite role by motivating Chinese applicants to make patent application,which is reflected by a rise of application and grant share by Chinese applicants in post-2001 period.Patent grant is viewed as one of the key indicators for judging the patent value.We find that foreign applicants from the U.S.,Japan and European Union have higher patent grant rate than that of Chinese applicants,suggesting that patents held by foreign applicants may have higher potential value than those held by Chinese applicants.Different types of applicants may differ from each other in terms of the patent grant rate,where research institutes have higher patent grant rate than that of corporations and individuals.Since the filed patents in China can enjoy a provisional protection for 3 years at most,some applicants do not request their examination.After an examination of the novelty,inventiveness and practicality of patent,the patent office determines whether to grant it.By using a bivariate probit model,we make an econometric analysis of this case.The result confirms the positive impact of both patent law amendments in 1993 and 2001 on the patent grant.We also find other determinants of the patent grant,for example,R&D human capital investment,applicant’s competitive technology advantage in the patent’s field,and applicant’s research ability,play positive roles in driving applicant to ask for examination.Patent’s complexity plays a positive role in driving the patent office to grant it.We discuss the significance of our study at the end of this paper.